xref: /freebsd/contrib/one-true-awk/FIXES (revision aa8b9e0706882484bdefbce15e3f0c4f1c049a63)
1/****************************************************************
2Copyright (C) Lucent Technologies 1997
3All Rights Reserved
4
5Permission to use, copy, modify, and distribute this software and
6its documentation for any purpose and without fee is hereby
7granted, provided that the above copyright notice appear in all
8copies and that both that the copyright notice and this
9permission notice and warranty disclaimer appear in supporting
10documentation, and that the name Lucent Technologies or any of
11its entities not be used in advertising or publicity pertaining
12to distribution of the software without specific, written prior
13permission.
14
15LUCENT DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE,
16INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS.
17IN NO EVENT SHALL LUCENT OR ANY OF ITS ENTITIES BE LIABLE FOR ANY
18SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
19WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER
20IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION,
21ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF
22THIS SOFTWARE.
23****************************************************************/
24
25This file lists all bug fixes, changes, etc., made since the AWK book
26was sent to the printers in August, 1987.
27
28May 6, 2011:
29	added #ifdef for isblank.
30	now allows -ffoo as well as -f foo arguments.
31	(thanks, ruslan)
32
33May 1, 2011:
34	after advice from todd miller, kevin lo, ruslan ermilov,
35	and arnold robbins, changed srand() to return the previous
36	seed (which is 1 on the first call of srand).  the seed is
37	an Awkfloat internally though converted to unsigned int to
38	pass to the library srand().  thanks, everyone.
39
40	fixed a subtle (and i hope low-probability) overflow error
41	in fldbld, by adding space for one extra \0.  thanks to
42	robert bassett for spotting this one and providing a fix.
43
44	removed the files related to compilation on windows.  i no
45	longer have anything like a current windows environment, so
46	i can't test any of it.
47
48May 23, 2010:
49	fixed long-standing overflow bug in run.c; many thanks to
50	nelson beebe for spotting it and providing the fix.
51
52	fixed bug that didn't parse -vd=1 properly; thanks to santiago
53	vila for spotting it.
54
55Feb 8, 2010:
56	i give up.  replaced isblank with isspace in b.c; there are
57	no consistent header files.
58
59Nov 26, 2009:
60	fixed a long-standing issue with when FS takes effect.  a
61	change to FS is now noticed immediately for subsequent splits.
62
63	changed the name getline() to awkgetline() to avoid yet another
64	name conflict somewhere.
65
66Feb 11, 2009:
67	temporarily for now defined HAS_ISBLANK, since that seems to
68	be the best way through the thicket.  isblank arrived in C99,
69	but seems to be arriving at different systems at different
70	times.
71
72Oct 8, 2008:
73	fixed typo in b.c that set tmpvec wrongly.  no one had ever
74	run into the problem, apparently.  thanks to alistair crooks.
75
76Oct 23, 2007:
77	minor fix in lib.c: increase inputFS to 100, change malloc
78	for fields to n+1.
79
80	fixed memory fault caused by out of order test in setsval.
81
82	thanks to david o'brien, freebsd, for both fixes.
83
84May 1, 2007:
85	fiddle in makefile to fix for BSD make; thanks to igor sobrado.
86
87Mar 31, 2007:
88	fixed some null pointer refs calling adjbuf.
89
90Feb 21, 2007:
91	fixed a bug in matching the null RE in sub and gsub.  thanks to al aho
92	who actually did the fix (in b.c), and to wolfgang seeberg for finding
93	it and providing a very compact test case.
94
95	fixed quotation in b.c; thanks to Hal Pratt and the Princeton Dante
96	Project.
97
98	removed some no-effect asserts in run.c.
99
100	fiddled maketab.c to not complain about bison-generated values.
101
102	removed the obsolete -V argument; fixed --version to print the
103	version and exit.
104
105	fixed wording and an outright error in the usage message; thanks to igor
106	sobrado and jason mcintyre.
107
108	fixed a bug in -d that caused core dump if no program followed.
109
110Jan 1, 2007:
111	dropped mac.code from makefile; there are few non-MacOSX
112	mac's these days.
113
114Jan 17, 2006:
115	system() not flagged as unsafe in the unadvertised -safe option.
116	found it while enhancing tests before shipping the ;login: article.
117	practice what you preach.
118
119	removed the 9-years-obsolete -mr and -mf flags.
120
121	added -version and --version options.
122
123	core dump on linux with BEGIN {nextfile}, now fixed.
124
125	removed some #ifdef's in run.c and lex.c that appear to no
126	longer be necessary.
127
128Apr 24, 2005:
129	modified lib.c so that values of $0 et al are preserved in the END
130	block, apparently as required by posix.  thanks to havard eidnes
131	for the report and code.
132
133Jan 14, 2005:
134	fixed infinite loop in parsing, originally found by brian tsang.
135	thanks to arnold robbins for a suggestion that started me
136	rethinking it.
137
138Dec 31, 2004:
139	prevent overflow of -f array in main, head off potential error in
140	call of SYNTAX(), test malloc return in lib.c, all with thanks to
141	todd miller.
142
143Dec 22, 2004:
144	cranked up size of NCHARS; coverity thinks it can be overrun with
145	smaller size, and i think that's right.  added some assertions to b.c
146	to catch places where it might overrun.  the RE code is still fragile.
147
148Dec 5, 2004:
149	fixed a couple of overflow problems with ridiculous field numbers:
150	e.g., print $(2^32-1).  thanks to ruslan ermilov, giorgos keramidas
151	and david o'brien at freebsd.org for patches.  this really should
152	be re-done from scratch.
153
154Nov 21, 2004:
155	fixed another 25-year-old RE bug, in split.  it's another failure
156	to (re-)initialize.  thanks to steve fisher for spotting this and
157	providing a good test case.
158
159Nov 22, 2003:
160	fixed a bug in regular expressions that dates (so help me) from 1977;
161	it's been there from the beginning.  an anchored longest match that
162	was longer than the number of states triggered a failure to initialize
163	the machine properly.  many thanks to moinak ghosh for not only finding
164	this one but for providing a fix, in some of the most mysterious
165	code known to man.
166
167	fixed a storage leak in call() that appears to have been there since
168	1983 or so -- a function without an explicit return that assigns a
169	string to a parameter leaked a Cell.  thanks to moinak ghosh for
170	spotting this very subtle one.
171
172Jul 31, 2003:
173	fixed, thanks to andrey chernov and ruslan ermilov, a bug in lex.c
174	that mis-handled the character 255 in input.  (it was being compared
175	to EOF with a signed comparison.)
176
177Jul 29, 2003:
178	fixed (i think) the long-standing botch that included the beginning of
179	line state ^ for RE's in the set of valid characters; this led to a
180	variety of odd problems, including failure to properly match certain
181	regular expressions in non-US locales.  thanks to ruslan for keeping
182	at this one.
183
184Jul 28, 2003:
185	n-th try at getting internationalization right, with thanks to volker
186	kiefel, arnold robbins and ruslan ermilov for advice, though they
187	should not be blamed for the outcome.  according to posix, "."  is the
188	radix character in programs and command line arguments regardless of
189	the locale; otherwise, the locale should prevail for input and output
190	of numbers.  so it's intended to work that way.
191
192	i have rescinded the attempt to use strcoll in expanding shorthands in
193	regular expressions (cclenter).  its properties are much too
194	surprising; for example [a-c] matches aAbBc in locale en_US but abBcC
195	in locale fr_CA.  i can see how this might arise by implementation
196	but i cannot explain it to a human user.  (this behavior can be seen
197	in gawk as well; we're leaning on the same library.)
198
199	the issue appears to be that strcoll is meant for sorting, where
200	merging upper and lower case may make sense (though note that unix
201	sort does not do this by default either).  it is not appropriate
202	for regular expressions, where the goal is to match specific
203	patterns of characters.  in any case, the notations [:lower:], etc.,
204	are available in awk, and they are more likely to work correctly in
205	most locales.
206
207	a moratorium is hereby declared on internationalization changes.
208	i apologize to friends and colleagues in other parts of the world.
209	i would truly like to get this "right", but i don't know what
210	that is, and i do not want to keep making changes until it's clear.
211
212Jul 4, 2003:
213	fixed bug that permitted non-terminated RE, as in "awk /x".
214
215Jun 1, 2003:
216	subtle change to split: if source is empty, number of elems
217	is always 0 and the array is not set.
218
219Mar 21, 2003:
220	added some parens to isblank, in another attempt to make things
221	internationally portable.
222
223Mar 14, 2003:
224	the internationalization changes, somewhat modified, are now
225	reinstated.  in theory awk will now do character comparisons
226	and case conversions in national language, but "." will always
227	be the decimal point separator on input and output regardless
228	of national language.  isblank(){} has an #ifndef.
229
230	this no longer compiles on windows: LC_MESSAGES isn't defined
231	in vc6++.
232
233	fixed subtle behavior in field and record splitting: if FS is
234	a single character and RS is not empty, \n is NOT a separator.
235	this tortuous reading is found in the awk book; behavior now
236	matches gawk and mawk.
237
238Dec 13, 2002:
239	for the moment, the internationalization changes of nov 29 are
240	rolled back -- programs like x = 1.2 don't work in some locales,
241	because the parser is expecting x = 1,2.  until i understand this
242	better, this will have to wait.
243
244Nov 29, 2002:
245	modified b.c (with tiny changes in main and run) to support
246	locales, using strcoll and iswhatever tests for posix character
247	classes.  thanks to ruslan ermilov (ru@freebsd.org) for code.
248	the function isblank doesn't seem to have propagated to any
249	header file near me, so it's there explicitly.  not properly
250	tested on non-ascii character sets by me.
251
252Jun 28, 2002:
253	modified run/format() and tran/getsval() to do a slightly better
254	job on using OFMT for output from print and CONVFMT for other
255	number->string conversions, as promised by posix and done by
256	gawk and mawk.  there are still places where it doesn't work
257	right if CONVFMT is changed; by then the STR attribute of the
258	variable has been irrevocably set.  thanks to arnold robbins for
259	code and examples.
260
261	fixed subtle bug in format that could get core dump.  thanks to
262	Jaromir Dolecek <jdolecek@NetBSD.org> for finding and fixing.
263	minor cleanup in run.c / format() at the same time.
264
265	added some tests for null pointers to debugging printf's, which
266	were never intended for external consumption.  thanks to dave
267	kerns (dkerns@lucent.com) for pointing this out.
268
269	GNU compatibility: an empty regexp matches anything (thanks to
270	dag-erling smorgrav, des@ofug.org).  subject to reversion if
271	this does more harm than good.
272
273	pervasive small changes to make things more const-correct, as
274	reported by gcc's -Wwrite-strings.  as it says in the gcc manual,
275	this may be more nuisance than useful.  provoked by a suggestion
276	and code from arnaud desitter, arnaud@nimbus.geog.ox.ac.uk
277
278	minor documentation changes to note that this now compiles out
279	of the box on Mac OS X.
280
281Feb 10, 2002:
282	changed types in posix chars structure to quiet solaris cc.
283
284Jan 1, 2002:
285	fflush() or fflush("") flushes all files and pipes.
286
287	length(arrayname) returns number of elements; thanks to
288	arnold robbins for suggestion.
289
290	added a makefile.win to make it easier to build on windows.
291	based on dan allen's buildwin.bat.
292
293Nov 16, 2001:
294	added support for posix character class names like [:digit:],
295	which are not exactly shorter than [0-9] and perhaps no more
296	portable.  thanks to dag-erling smorgrav for code.
297
298Feb 16, 2001:
299	removed -m option; no longer needed, and it was actually
300	broken (noted thanks to volker kiefel).
301
302Feb 10, 2001:
303	fixed an appalling bug in gettok: any sequence of digits, +,-, E, e,
304	and period was accepted as a valid number if it started with a period.
305	this would never have happened with the lex version.
306
307	other 1-character botches, now fixed, include a bare $ and a
308	bare " at the end of the input.
309
310Feb 7, 2001:
311	more (const char *) casts in b.c and tran.c to silence warnings.
312
313Nov 15, 2000:
314	fixed a bug introduced in august 1997 that caused expressions
315	like $f[1] to be syntax errors.  thanks to arnold robbins for
316	noticing this and providing a fix.
317
318Oct 30, 2000:
319	fixed some nextfile bugs: not handling all cases.  thanks to
320	arnold robbins for pointing this out.  new regressions added.
321
322	close() is now a function.  it returns whatever the library
323	fclose returns, and -1 for closing a file or pipe that wasn't
324	opened.
325
326Sep 24, 2000:
327	permit \n explicitly in character classes; won't work right
328	if comes in as "[\n]" but ok as /[\n]/, because of multiple
329	processing of \'s.  thanks to arnold robbins.
330
331July 5, 2000:
332	minor fiddles in tran.c to keep compilers happy about uschar.
333	thanks to norman wilson.
334
335May 25, 2000:
336	yet another attempt at making 8-bit input work, with another
337	band-aid in b.c (member()), and some (uschar) casts to head
338	off potential errors in subscripts (like isdigit).  also
339	changed HAT to NCHARS-2.  thanks again to santiago vila.
340
341	changed maketab.c to ignore apparently out of range definitions
342	instead of halting; new freeBSD generates one.  thanks to
343	jon snader <jsnader@ix.netcom.com> for pointing out the problem.
344
345May 2, 2000:
346	fixed an 8-bit problem in b.c by making several char*'s into
347	unsigned char*'s.  not clear i have them all yet.  thanks to
348	Santiago Vila <sanvila@unex.es> for the bug report.
349
350Apr 21, 2000:
351	finally found and fixed a memory leak in function call; it's
352	been there since functions were added ~1983.  thanks to
353	jon bentley for the test case that found it.
354
355	added test in envinit to catch environment "variables" with
356	names beginning with '='; thanks to Berend Hasselman.
357
358Jul 28, 1999:
359	added test in defn() to catch function foo(foo), which
360	otherwise recurses until core dump.  thanks to arnold
361	robbins for noticing this.
362
363Jun 20, 1999:
364	added *bp in gettok in lex.c; appears possible to exit function
365	without terminating the string.  thanks to russ cox.
366
367Jun 2, 1999:
368	added function stdinit() to run to initialize files[] array,
369	in case stdin, etc., are not constants; some compilers care.
370
371May 10, 1999:
372	replaced the ERROR ... FATAL, etc., macros with functions
373	based on vprintf, to avoid problems caused by overrunning
374	fixed-size errbuf array.  thanks to ralph corderoy for the
375	impetus, and for pointing out a string termination bug in
376	qstring as well.
377
378Apr 21, 1999:
379	fixed bug that caused occasional core dumps with commandline
380	variable with value ending in \.  (thanks to nelson beebe for
381	the test case.)
382
383Apr 16, 1999:
384	with code kindly provided by Bruce Lilly, awk now parses
385	/=/ and similar constructs more sensibly in more places.
386	Bruce also provided some helpful test cases.
387
388Apr 5, 1999:
389	changed true/false to True/False in run.c to make it
390	easier to compile with C++.  Added some casts on malloc
391	and realloc to be honest about casts; ditto.  changed
392	ltype int to long in struct rrow to reduce some 64-bit
393	complaints; other changes scattered throughout for the
394	same purpose.  thanks to Nelson Beebe for these portability
395	improvements.
396
397	removed some horrible pointer-int casting in b.c and elsewhere
398	by adding ptoi and itonp to localize the casts, which are
399	all benign.  fixed one incipient bug that showed up on sgi
400	in 64-bit mode.
401
402	reset lineno for new source file; include filename in error
403	message.  also fixed line number error in continuation lines.
404	(thanks to Nelson Beebe for both of these.)
405
406Mar 24, 1999:
407	Nelson Beebe notes that irix 5.3 yacc dies with a bogus
408	error; use a newer version or switch to bison, since sgi
409	is unlikely to fix it.
410
411Mar 5, 1999:
412	changed isnumber to is_number to avoid the problem caused by
413	versions of ctype.h that include the name isnumber.
414
415	distribution now includes a script for building on a Mac,
416	thanks to Dan Allen.
417
418Feb 20, 1999:
419	fixed memory leaks in run.c (call) and tran.c (setfval).
420	thanks to Stephen Nutt for finding these and providing the fixes.
421
422Jan 13, 1999:
423	replaced srand argument by (unsigned int) in run.c;
424	avoids problem on Mac and potentially on Unix & Windows.
425	thanks to Dan Allen.
426
427	added a few (int) casts to silence useless compiler warnings.
428	e.g., errorflag= in run.c jump().
429
430	added proctab.c to the bundle outout; one less thing
431	to have to compile out of the box.
432
433	added calls to _popen and _pclose to the win95 stub for
434	pipes (thanks to Steve Adams for this helpful suggestion).
435	seems to work, though properties are not well understood
436	by me, and it appears that under some circumstances the
437	pipe output is truncated.  Be careful.
438
439Oct 19, 1998:
440	fixed a couple of bugs in getrec: could fail to update $0
441	after a getline var; because inputFS wasn't initialized,
442	could split $0 on every character, a misleading diversion.
443
444	fixed caching bug in makedfa: LRU was actually removing
445	least often used.
446
447	thanks to ross ridge for finding these, and for providing
448	great bug reports.
449
450May 12, 1998:
451	fixed potential bug in readrec: might fail to update record
452	pointer after growing.  thanks to dan levy for spotting this
453	and suggesting the fix.
454
455Mar 12, 1998:
456	added -V to print version number and die.
457
458Feb 11, 1998:
459	subtle silent bug in lex.c: if the program ended with a number
460	longer than 1 digit, part of the input would be pushed back and
461	parsed again because token buffer wasn't terminated right.
462	example:  awk 'length($0) > 10'.  blush.  at least i found it
463	myself.
464
465Aug 31, 1997:
466	s/adelete/awkdelete/: SGI uses this in malloc.h.
467	thanks to nelson beebe for pointing this one out.
468
469Aug 21, 1997:
470	fixed some bugs in sub and gsub when replacement includes \\.
471	this is a dark, horrible corner, but at least now i believe that
472	the behavior is the same as gawk and the intended posix standard.
473	thanks to arnold robbins for advice here.
474
475Aug 9, 1997:
476	somewhat regretfully, replaced the ancient lex-based lexical
477	analyzer with one written in C.  it's longer, generates less code,
478	and more portable; the old one depended too much on mysterious
479	properties of lex that were not preserved in other environments.
480	in theory these recognize the same language.
481
482	now using strtod to test whether a string is a number, instead of
483	the convoluted original function.  should be more portable and
484	reliable if strtod is implemented right.
485
486	removed now-pointless optimization in makefile that tries to avoid
487	recompilation when awkgram.y is changed but symbols are not.
488
489	removed most fixed-size arrays, though a handful remain, some
490	of which are unchecked.  you have been warned.
491
492Aug 4, 1997:
493	with some trepidation, replaced the ancient code that managed
494	fields and $0 in fixed-size arrays with arrays that grow on
495	demand.  there is still some tension between trying to make this
496	run fast and making it clean; not sure it's right yet.
497
498	the ill-conceived -mr and -mf arguments are now useful only
499	for debugging.  previous dynamic string code removed.
500
501	numerous other minor cleanups along the way.
502
503Jul 30, 1997:
504	using code provided by dan levy (to whom profuse thanks), replaced
505	fixed-size arrays and awkward kludges by a fairly uniform mechanism
506	to grow arrays as needed for printf, sub, gsub, etc.
507
508Jul 23, 1997:
509	falling off the end of a function returns "" and 0, not 0.
510	thanks to arnold robbins.
511
512Jun 17, 1997:
513	replaced several fixed-size arrays by dynamically-created ones
514	in run.c; added overflow tests to some previously unchecked cases.
515	getline, toupper, tolower.
516
517	getline code is still broken in that recursive calls may wind
518	up using the same space.  [fixed later]
519
520	increased RECSIZE to 8192 to push problems further over the horizon.
521
522	added \r to \n as input line separator for programs, not data.
523	damn CRLFs.
524
525	modified format() to permit explicit printf("%c", 0) to include
526	a null byte in output.  thanks to ken stailey for the fix.
527
528	added a "-safe" argument that disables file output (print >,
529	print >>), process creation (cmd|getline, print |, system), and
530	access to the environment (ENVIRON).  this is a first approximation
531	to a "safe" version of awk, but don't rely on it too much.  thanks
532	to joan feigenbaum and matt blaze for the inspiration long ago.
533
534Jul 8, 1996:
535	fixed long-standing bug in sub, gsub(/a/, "\\\\&"); thanks to
536	ralph corderoy.
537
538Jun 29, 1996:
539	fixed awful bug in new field splitting; didn't get all the places
540	where input was done.
541
542Jun 28, 1996:
543	changed field-splitting to conform to posix definition: fields are
544	split using the value of FS at the time of input; it used to be
545	the value when the field or NF was first referred to, a much less
546	predictable definition.  thanks to arnold robbins for encouragement
547	to do the right thing.
548
549May 28, 1996:
550	fixed appalling but apparently unimportant bug in parsing octal
551	numbers in reg exprs.
552
553	explicit hex in reg exprs now limited to 2 chars: \xa, \xaa.
554
555May 27, 1996:
556	cleaned up some declarations so gcc -Wall is now almost silent.
557
558	makefile now includes backup copies of ytab.c and lexyy.c in case
559	one makes before looking; it also avoids recreating lexyy.c unless
560	really needed.
561
562	s/aprintf/awkprint, s/asprintf/awksprintf/ to avoid some name clashes
563	with unwisely-written header files.
564
565	thanks to jeffrey friedl for several of these.
566
567May 26, 1996:
568	an attempt to rationalize the (unsigned) char issue.  almost all
569	instances of unsigned char have been removed; the handful of places
570	in b.c where chars are used as table indices have been hand-crafted.
571	added some latin-1 tests to the regression, but i'm not confident;
572	none of my compilers seem to care much.  thanks to nelson beebe for
573	pointing out some others that do care.
574
575May 2, 1996:
576	removed all register declarations.
577
578	enhanced split(), as in gawk, etc:  split(s, a, "") splits s into
579	a[1]...a[length(s)] with each character a single element.
580
581	made the same changes for field-splitting if FS is "".
582
583	added nextfile, as in gawk: causes immediate advance to next
584	input file. (thanks to arnold robbins for inspiration and code).
585
586	small fixes to regexpr code:  can now handle []], [[], and
587	variants;  [] is now a syntax error, rather than matching
588	everything;  [z-a] is now empty, not z.  far from complete
589	or correct, however.  (thanks to jeffrey friedl for pointing out
590	some awful behaviors.)
591
592Apr 29, 1996:
593	replaced uchar by uschar everywhere; apparently some compilers
594	usurp this name and this causes conflicts.
595
596	fixed call to time in run.c (bltin); arg is time_t *.
597
598	replaced horrible pointer/long punning in b.c by a legitimate
599	union.  should be safer on 64-bit machines and cleaner everywhere.
600	(thanks to nelson beebe for pointing out some of these problems.)
601
602	replaced nested comments by #if 0...#endif in run.c, lib.c.
603
604	removed getsval, setsval, execute macros from run.c and lib.c.
605	machines are 100x faster than they were when these macros were
606	first used.
607
608	revised filenames: awk.g.y => awkgram.y, awk.lx.l => awklex.l,
609	y.tab.[ch] => ytab.[ch], lex.yy.c => lexyy.c, all in the aid of
610	portability to nameless systems.
611
612	"make bundle" now includes yacc and lex output files for recipients
613	who don't have yacc or lex.
614
615Aug 15, 1995:
616	initialized Cells in setsymtab more carefully; some fields
617	were not set.  (thanks to purify, all of whose complaints i
618	think i now understand.)
619
620	fixed at least one error in gsub that looked at -1-th element
621	of an array when substituting for a null match (e.g., $).
622
623	delete arrayname is now legal; it clears the elements but leaves
624	the array, which may not be the right behavior.
625
626	modified makefile: my current make can't cope with the test used
627	to avoid unnecessary yacc invocations.
628
629Jul 17, 1995:
630	added dynamically growing strings to awk.lx.l and b.c
631	to permit regular expressions to be much bigger.
632	the state arrays can still overflow.
633
634Aug 24, 1994:
635	detect duplicate arguments in function definitions (mdm).
636
637May 11, 1994:
638	trivial fix to printf to limit string size in sub().
639
640Apr 22, 1994:
641	fixed yet another subtle self-assignment problem:
642	$1 = $2; $1 = $1 clobbered $1.
643
644	Regression tests now use private echo, to avoid quoting problems.
645
646Feb 2, 1994:
647	changed error() to print line number as %d, not %g.
648
649Jul 23, 1993:
650	cosmetic changes: increased sizes of some arrays,
651	reworded some error messages.
652
653	added CONVFMT as in posix (just replaced OFMT in getsval)
654
655	FILENAME is now "" until the first thing that causes a file
656	to be opened.
657
658Nov 28, 1992:
659	deleted yyunput and yyoutput from proto.h;
660	different versions of lex give these different declarations.
661
662May 31, 1992:
663	added -mr N and -mf N options: more record and fields.
664	these really ought to adjust automatically.
665
666	cleaned up some error messages; "out of space" now means
667	malloc returned NULL in all cases.
668
669	changed rehash so that if it runs out, it just returns;
670	things will continue to run slow, but maybe a bit longer.
671
672Apr 24, 1992:
673	remove redundant close of stdin when using -f -.
674
675	got rid of core dump with -d; awk -d just prints date.
676
677Apr 12, 1992:
678	added explicit check for /dev/std(in,out,err) in redirection.
679	unlike gawk, no /dev/fd/n yet.
680
681	added (file/pipe) builtin.  hard to test satisfactorily.
682	not posix.
683
684Feb 20, 1992:
685	recompile after abortive changes;  should be unchanged.
686
687Dec 2, 1991:
688	die-casting time:  converted to ansi C, installed that.
689
690Nov 30, 1991:
691	fixed storage leak in freefa, failing to recover [N]CCL.
692	thanks to Bill Jones (jones@cs.usask.ca)
693
694Nov 19, 1991:
695	use RAND_MAX instead of literal in builtin().
696
697Nov 12, 1991:
698	cranked up some fixed-size arrays in b.c, and added a test for
699	overflow in penter.  thanks to mark larsen.
700
701Sep 24, 1991:
702	increased buffer in gsub.  a very crude fix to a general problem.
703	and again on Sep 26.
704
705Aug 18, 1991:
706	enforce variable name syntax for commandline variables: has to
707	start with letter or _.
708
709Jul 27, 1991:
710	allow newline after ; in for statements.
711
712Jul 21, 1991:
713	fixed so that in self-assignment like $1=$1, side effects
714	like recomputing $0 take place.  (this is getting subtle.)
715
716Jun 30, 1991:
717	better test for detecting too-long output record.
718
719Jun 2, 1991:
720	better defense against very long printf strings.
721	made break and continue illegal outside of loops.
722
723May 13, 1991:
724	removed extra arg on gettemp, tempfree.  minor error message rewording.
725
726May 6, 1991:
727	fixed silly bug in hex parsing in hexstr().
728	removed an apparently unnecessary test in isnumber().
729	warn about weird printf conversions.
730	fixed unchecked array overwrite in relex().
731
732	changed for (i in array) to access elements in sorted order.
733	then unchanged it -- it really does run slower in too many cases.
734	left the code in place, commented out.
735
736Feb 10, 1991:
737	check error status on all writes, to avoid banging on full disks.
738
739Jan 28, 1991:
740	awk -f - reads the program from stdin.
741
742Jan 11, 1991:
743	failed to set numeric state on $0 in cmd|getline context in run.c.
744
745Nov 2, 1990:
746	fixed sleazy test for integrality in getsval;  use modf.
747
748Oct 29, 1990:
749	fixed sleazy buggy code in lib.c that looked (incorrectly) for
750	too long input lines.
751
752Oct 14, 1990:
753	fixed the bug on p. 198 in which it couldn't deduce that an
754	argument was an array in some contexts.  replaced the error
755	message in intest() by code that damn well makes it an array.
756
757Oct 8, 1990:
758	fixed horrible bug:  types and values were not preserved in
759	some kinds of self-assignment. (in assign().)
760
761Aug 24, 1990:
762	changed NCHARS to 256 to handle 8-bit characters in strings
763	presented to match(), etc.
764
765Jun 26, 1990:
766	changed struct rrow (awk.h) to use long instead of int for lval,
767	since cfoll() stores a pointer in it.  now works better when int's
768	are smaller than pointers!
769
770May 6, 1990:
771	AVA fixed the grammar so that ! is uniformly of the same precedence as
772	unary + and -.  This renders illegal some constructs like !x=y, which
773	now has to be parenthesized as !(x=y), and makes others work properly:
774	!x+y is (!x)+y, and x!y is x !y, not two pattern-action statements.
775	(These problems were pointed out by Bob Lenk of Posix.)
776
777	Added \x to regular expressions (already in strings).
778	Limited octal to octal digits; \8 and \9 are not octal.
779	Centralized the code for parsing escapes in regular expressions.
780	Added a bunch of tests to T.re and T.sub to verify some of this.
781
782Feb 9, 1990:
783	fixed null pointer dereference bug in main.c:  -F[nothing].  sigh.
784
785	restored srand behavior:  it returns the current seed.
786
787Jan 18, 1990:
788	srand now returns previous seed value (0 to start).
789
790Jan 5, 1990:
791	fix potential problem in tran.c -- something was freed,
792	then used in freesymtab.
793
794Oct 18, 1989:
795	another try to get the max number of open files set with
796	relatively machine-independent code.
797
798	small fix to input() in case of multiple reads after EOF.
799
800Oct 11, 1989:
801	FILENAME is now defined in the BEGIN block -- too many old
802	programs broke.
803
804	"-" means stdin in getline as well as on the commandline.
805
806	added a bunch of casts to the code to tell the truth about
807	char * vs. unsigned char *, a right royal pain.  added a
808	setlocale call to the front of main, though probably no one
809	has it usefully implemented yet.
810
811Aug 24, 1989:
812	removed redundant relational tests against nullnode if parse
813	tree already had a relational at that point.
814
815Aug 11, 1989:
816	fixed bug:  commandline variable assignment has to look like
817	var=something.  (consider the man page for =, in file =.1)
818
819	changed number of arguments to functions to static arrays
820	to avoid repeated malloc calls.
821
822Aug 2, 1989:
823	restored -F (space) separator
824
825Jul 30, 1989:
826	added -v x=1 y=2 ... for immediate commandline variable assignment;
827	done before the BEGIN block for sure.  they have to precede the
828	program if the program is on the commandline.
829	Modified Aug 2 to require a separate -v for each assignment.
830
831Jul 10, 1989:
832	fixed ref-thru-zero bug in environment code in tran.c
833
834Jun 23, 1989:
835	add newline to usage message.
836
837Jun 14, 1989:
838	added some missing ansi printf conversion letters: %i %X %E %G.
839	no sensible meaning for h or L, so they may not do what one expects.
840
841	made %* conversions work.
842
843	changed x^y so that if n is a positive integer, it's done
844	by explicit multiplication, thus achieving maximum accuracy.
845	(this should be done by pow() but it seems not to be locally.)
846	done to x ^= y as well.
847
848Jun 4, 1989:
849	ENVIRON array contains environment: if shell variable V=thing,
850		ENVIRON["V"] is "thing"
851
852	multiple -f arguments permitted.  error reporting is naive.
853	(they were permitted before, but only the last was used.)
854
855	fixed a really stupid botch in the debugging macro dprintf
856
857	fixed order of evaluation of commandline assignments to match
858	what the book claims:  an argument of the form x=e is evaluated
859	at the time it would have been opened if it were a filename (p 63).
860	this invalidates the suggested answer to ex 4-1 (p 195).
861
862	removed some code that permitted -F (space) fieldseparator,
863	since it didn't quite work right anyway.  (restored aug 2)
864
865Apr 27, 1989:
866	Line number now accumulated correctly for comment lines.
867
868Apr 26, 1989:
869	Debugging output now includes a version date,
870	if one compiles it into the source each time.
871
872Apr 9, 1989:
873	Changed grammar to prohibit constants as 3rd arg of sub and gsub;
874	prevents class of overwriting-a-constant errors.  (Last one?)
875	This invalidates the "banana" example on page 43 of the book.
876
877	Added \a ("alert"), \v (vertical tab), \xhhh (hexadecimal),
878	as in ANSI, for strings.  Rescinded the sloppiness that permitted
879	non-octal digits in \ooo.  Warning:  not all compilers and libraries
880	will be able to deal with \x correctly.
881
882Jan 9, 1989:
883	Fixed bug that caused tempcell list to contain a duplicate.
884	The fix is kludgy.
885
886Dec 17, 1988:
887	Catches some more commandline errors in main.
888	Removed redundant decl of modf in run.c (confuses some compilers).
889	Warning:  there's no single declaration of malloc, etc., in awk.h
890	that seems to satisfy all compilers.
891
892Dec 7, 1988:
893	Added a bit of code to error printing to avoid printing nulls.
894	(Not clear that it actually would.)
895
896Nov 27, 1988:
897	With fear and trembling, modified the grammar to permit
898	multiple pattern-action statements on one line without
899	an explicit separator.  By definition, this capitulation
900	to the ghost of ancient implementations remains undefined
901	and thus subject to change without notice or apology.
902	DO NOT COUNT ON IT.
903
904Oct 30, 1988:
905	Fixed bug in call() that failed to recover storage.
906
907	A warning is now generated if there are more arguments
908	in the call than in the definition (in lieu of fixing
909	another storage leak).
910
911Oct 20, 1988:
912	Fixed %c:  if expr is numeric, use numeric value;
913	otherwise print 1st char of string value.  still
914	doesn't work if the value is 0 -- won't print \0.
915
916	Added a few more checks for running out of malloc.
917
918Oct 12, 1988:
919	Fixed bug in call() that freed local arrays twice.
920
921	Fixed to handle deletion of non-existent array right;
922	complains about attempt to delete non-array element.
923
924Sep 30, 1988:
925	Now guarantees to evaluate all arguments of built-in
926	functions, as in C;  the appearance is that arguments
927	are evaluated before the function is called.  Places
928	affected are sub (gsub was ok), substr, printf, and
929	all the built-in arithmetic functions in bltin().
930	A warning is generated if a bltin() is called with
931	the wrong number of arguments.
932
933	This requires changing makeprof on p167 of the book.
934
935Aug 23, 1988:
936	setting FILENAME in BEGIN caused core dump, apparently
937	because it was freeing space not allocated by malloc.
938
939July 24, 1988:
940	fixed egregious error in toupper/tolower functions.
941	still subject to rescinding, however.
942
943July 2, 1988:
944	flush stdout before opening file or pipe
945
946July 2, 1988:
947	performance bug in b.c/cgoto(): not freeing some sets of states.
948	partial fix only right now, and the number of states increased
949	to make it less obvious.
950
951June 1, 1988:
952	check error status on close
953
954May 28, 1988:
955	srand returns seed value it's using.
956	see 1/18/90
957
958May 22, 1988:
959	Removed limit on depth of function calls.
960
961May 10, 1988:
962	Fixed lib.c to permit _ in commandline variable names.
963
964Mar 25, 1988:
965	main.c fixed to recognize -- as terminator of command-
966	line options.  Illegal options flagged.
967	Error reporting slightly cleaned up.
968
969Dec 2, 1987:
970	Newer C compilers apply a strict scope rule to extern
971	declarations within functions.  Two extern declarations in
972	lib.c and tran.c have been moved to obviate this problem.
973
974Oct xx, 1987:
975	Reluctantly added toupper and tolower functions.
976	Subject to rescinding without notice.
977
978Sep 17, 1987:
979	Error-message printer had printf(s) instead of
980	printf("%s",s);  got core dumps when the message
981	included a %.
982
983Sep 12, 1987:
984	Very long printf strings caused core dump;
985	fixed aprintf, asprintf, format to catch them.
986	Can still get a core dump in printf itself.
987
988
989